Recognition of Sleep Apnea by EEG Using Nonlinear Dynamics Methods

Obstructive sleep apnea is a disorder characterized by frequent interruptions in breathing during sleep, usually accompanied by snoring. Currently, obstructive sleep apnea syndrome is diagnosed using polysomnography, a method of recording body activity during sleep and pulse oximetry, which measures the amount of oxygen in the blood at any time. Obstructive sleep apnea itself is not a life-threatening condition, but it can cause serious problems such as cardiovascular and cerebrovascular disease. Sleep apnea during the day can make a person feel excessively sleepy, have difficulty concentrating, or headaches. Electroencephalogram (EEG) signals were used for the study. Signals are taken from the CAP Sleep database [1]. Sleep apnea has been associated with disorders such as narcolepsy, bruxism, periodic limb movement syndrome, sleep apnea, and nocturnal frontal frontal lobe epilepsy. It should be noted that these disorders are found in different sections of the ICSD-3 classification. The paper investigates entropy methods and the calculation of the highest Lyapunov exponent in relation to the problem of recognizing sleep apnea by EEG. The developed algorithms and programs are intended for medical computer systems and implemented in the MATLAB software environment.